Monday, November 20, 2017

Distortion by the Numbers

John Sau has posted about how low THD+N should be to absolutely guarantee it is inaudible.

If the system reaches 100dB peaks, we can guarantee that the distortion will be inaudible if it is -100dB, because 0dB is the ultimate limit of hearing sensitivity.  Regardless of masking.  (Actually, in many cases, distortion is masked at much higher levels).

I like this.  I expect many of my numbers-phobic audiophile friends will not.

However, distortion meeting the astronomically small numbers recommended may be inarguably inaudible, however it is far less distortion than is required to achieve basic inaudibility virtually all of the time.

With devices which achieve 0.03% THD, and -110 to -120dB below peak noise (and peak...is an attainable level in my system), my system does not quite meet the recommendation actually, but I believe is adequate, as far as electronics are concerned.

The Acoustats are low distortion compared to typical box speakers, but perhaps not quite the lowest distortion in the lowest distortion speakers of all time.  They have more distortion in the bass than elsewhere, since I cross over at 100 Hz at 24dB/octave, that bass distortion has slight effect, and the rest is well below 1% if not 0.1%.  Still, in any fair assessment, the speakers generate the most distortion in my system, if not most systems, except for some tweako ones.



Monday, November 13, 2017

Amirm is back!

I was saddened to see that Amirm was bounced out of two other websites I have enjoyed.  He, and his purely objectivist interlocutor Ethan Winer, were two of the most interesting people at WhatsBestForum, for example, and now they've both been bounced.

I wouldn't say I would agree with him always, or even a lot, but he's interesting to read and one of the better ones at combining subjectivism with some kind of objectivism.  About as entertaining and probably at least as accurate as J Peter Montcrief--who doesn't seem to have written anything in a long time.

Well, now he has founded a new audio website himself, and it looks good enough to be added to my sidebar (actually, some of those you might think would have been taken off long ago, but whatever, I like to tease, just like these guys do).



Looks like it's filled with refugees from WBF and the like.

Tuesday, November 7, 2017

Another Amplifier ???

I saw a very nice looking Hafler 9300 just listed on eBay for $395.  I took a look at this review

http://www.hifi-classic.net/review/hafler-9300-120.html

They measured distortion of 0.0025% at 1 watt, the most important watt, where it was continuing to decrease.  And bandwidth over 200 kHz.  Rated distortion is 0.02% at 150 watts into 8 ohms, and 225 watts into 4 ohms.  They measured over 400 watts of dynamic power into 4 ohms.

OK, so I know most subjectivist audiophiles aren't impressed by such measurements, but they are very impressive to me.

It is of course the Jim Strickland Trans Nova design, with help from David Hafler.  Two of the greatest designers in the history of audio, and this (along with the Hafler 9500 and 9505) are the last Trans Nova amplifiers built by the Hafler company.  Hafler was a longtime friend of Strickland and he loved the Trans Nova design, a big part of the reason why he bought Acoustat.

Fosgate rebranded the Hafler 9505 as a Fosgate model after they bought Hafler.  Another endorsement.

All the Trans Nova amps will drive anything, and especially Acoustats of course which are about the hardest "4 ohm" speaker to drive ever made.  But these are the last Trans Nova amplifiers, and probably the best.

The 9500 and 9505 are the big versions, with about double the power.  They command significantly higher prices.  The 9500 got a rave review Stereophile, and had stellar measurements also.

Then I looked at the schematic, and knew instantly I had to have this amplifier.  It is simple, and perfect!  Check this out:

3 stages, can't get lower than that without compromises.

Dual diff amp JFETs in front.  The optimum circuit.

Bipolar drivers.  Also the optimum.  Substantial current gain is required to drive the capacitance of the output stage.

MOSFET outputs in the Trans Nova design that drives anything, with no protection circuitry needed.

My friend Tim was blown away as well.  He suggested it could well be my best sounding amplifier.  He noted the very low feedback (remember the rule, you either want very little or lots and lots...this is the very little side).  He noted the lack of emitter resistors on the outputs.  Normally those are required for thermal stability...but because they add additional feedback they actually create a bit more distortion.

Somehow this amp gets away without the emitter resistors, but still drives anything, with low distortion.

A miracle that hardly anyone noticed.  Except that if you search for reviews by people who have owned 9300 or 9505, they say it's the best they've ever heard.  People puts lots of effort into refurbing them (the low profile capacitors were custom made and can only be substituted with more smaller caps on a circuit board stuffed in) and then say afterwards, it was well worth it, the amp is so wonderful.

So now I have a simple amplifier to try.  Probably as good as any by Pass, or maybe better.

This is not your father's Hafler DH300, a good but fairly standard amplifier.

This looks like it may be the crown jewel of simple yet capable amplifier designs.

And, it must be great for driving Acoustats.  Some of the Pass designs are not, especially the simpler ones.


*** Update: Measurements

My unit is mint-like, came in factory box without any visible issues.  A chunky amp just slightly bigger than Parasound HCA-1000A, with nice rounded-edge heatsinks that look and work very well, and weighing quite a bit more, feels like a brick.

I had no scope hooked up, just relied on readings from my ST-1700B, which I discovered starts malfunctioning after about 3 hours.  I deliberately tuned it to work best before 3 hours, so I wouldn't have to wait that long.  Back then, a decade or so ago, it still worked OK after 3 hours.  Now it doesn't so much, and at 24 hours...it is hopeless, all distortion measurements are 100%.

Anyway, it seemed output could reach to about 60V into 8 ohms, that's 450W peak.  I measured at 30V, which is 112.5W, not quite rated power of 150W/8 ohms, and then did one measurement at 40V, which is 200W, which is above rated power.  At 40V, 1kHz, distortion is 0.045% L...better than many if not most amplifiiers within their ratings, but slightly worse still respectable at 0.16% R.

At 30V, distortion (THD+N, residual 0.0028%) was .005% L and 0.0055% R at 1kHz, .005% L and .0075% R at 20 Hz, .029%L and .05%R at 10kHz, and .067% L and 0.08% R at 20kHz.  The 20kHz distortion doesn't quite meet spec, but is nearly as impressive as everything else.  This is outstanding technical performance, the best I've ever measured in a power amplifier, and nearing my residual at 1kHz and 20 Hz (!).  There may be some aging effect as the amplifier is 25 years old, possibly explaining the out-of-speck 20kHz performance.

Stellar performance...AND it is a very simple design with low feedback.  An incredible "sleeper."  Great peak power, high power bandwidth (actually...I did not measure power bandwidth).

Offset a tad high, at 28mV L and 31mV R.  Perhaps some kind of aging has affected both the high frequency distortion, which is a bit out of spec, as well as the offset.  Still, few amplifiers have lower distortion at 20kHz, and the offset is tolerable for a power amplifier and low impedance speakers.

Idles to 114F to 120F on the heatsinks, less toward the back.  This is notably but not dangerously warm, and also shows the rather high bias operation, even with heatsinks which look to be larger than those on the equivalently powerful Parasound HCA 1500A.

This does look like it would be second to none on the Acoustats, except possibly in dynamics as it is unlikely it has power stiffness as the Krell, or quite as much output at 4 ohms.

But, like the Aragon, there is no limiting circuitry to cause the amp to shut down.  It just keeps on swinging.